Dysfunctional breathing in long COVID: inter-rater agreement and clinical characteristics

Abstract Background Persistent symptoms such as dyspnea and exercise intolerance are frequently reported in patients with Long COVID. Cardiopulmonary exercise testing (CPET) has emerged as an essential assessment to identify the etiology of these symptoms. Several studies on Long COVID have documented reductions in peak oxygen uptake (VO2peak) as well as abnormalities in other relevant CPET variables. An intriguing and not yet fully understood finding during a CPET assessment is the presence of dysfunctional breathing (DB). Little is known about its prevalence, clinical characteristics, and inter-rater agreement for its identification, which is aggravated by the absence of standardized criteria for diagnosis. Purpose The purpose of this study is to analyze the inter-rater agreement in identifying DB during CPET in patients with Long COVID and to evaluate the clinical characteristics of patients with DB. Methods In this observational study, CPET was performed on Long COVID patients, and the identification of DB was done by two independent raters. The inter-rater agreement was calculated using Cohen's Kappa coefficient, as well as the proportion of absolute and specific agreement. The clinical characteristics of patients with DB (defined by double agreement) were described, and the differences between subgroups were assessed using the Mann-Whitney test or Chi-Square test. Results The sample consisted of 109 patients (51% males). The median age was 54 years interquartile range (IQR): 42.5, 63.5. Critical illness was most prevalent during the acute phase (57%), followed by severe (24%) and moderate (19%). Dyspnea as a residual symptom was reported by 69 individuals (65%). In CPET assessments, the median VO2peak was 17.8 mL/kg/min (IQR: 14.5, 23.0), representing 58.9% (IQR: 49.9, 73.9) of the predicted values (Figure 1). Rater 1 identified 63 cases with DB (58%), and Rater 2 identified 48 cases (44%). Positive double agreement (confirmed cases of DB) resulted in 39 patients (36%) (Figure 2). The overall agreement between the raters was 70% (76 out of 109 cases), and the specific agreement (presence of DB) was 54% (39 out of 72 cases). The kappa was 0.41, indicating moderate agreement. Regarding clinical characteristics, the subgroup with DB had a higher prevalence of females, along with lower median VO2peak 16.3 (IQR: 13.9, 19.1) vs. 19.4 (IQR: 15.2, 23.8); p = 0.025, as well as lower peak heart rate and lower peak ventilation. However, ventilatory reserve showed no significant difference. There were no statistical differences in the severity of COVID-19 or in the presence of symptoms between the subgroups (Figure 1). Conclusion This study revealed a relatively high prevalence of DB in Long COVID patients referred for CPET assessment. The moderate inter-rater agreement underscores the diagnostic challenge posed by this abnormality during CPET assessments. Patients with the presence of DB were mainly females and had lower cardiorespiratory fitness.